Friday, 13 May 2011

Risk, season of birth and autism

Risk is a strange thing. We are told by many different sources about the risk of this happening, or that happening, normally accompanied by some smart figure. This page offers quite a good example of risk in action based on the UK mortality figures. What the statistics tell me is that, as a man, I am at greater risk than a woman of death at every stage of my life, from my first breath to my last, and without giving my age to you, I am currently on the borderline of a low-to-moderate risk of dying anytime soon. Great (he says reaching for the Smiths CD).

I was in two minds whether or not to make this post relating to possible seasonal influences and risk of receiving a diagnosis of an autism spectrum condition. Two minds because you could give me almost any reasonably well-known condition and I could probably provide some evidence that season or month of birth is potentially connected in some way. For example, the risk of brain tumor is greater for those born in the Winter according to this study. Interesting yes? But consider how many factors might be involved in developing a brain tumor - not least your subsequent exposure events to lots of things - and then such statistics begin to mean less and less.

I am not disputing that there may be a connection between some conditions and season of birth. There are perhaps some important connections to be made in respect to things like maternal risk of exposure to viral or other infectious agents, whether pollen counts are high, sunlight exposure, and even depending on where you live, whether or not there is increased exposure to environmental pollutants such as car fumes or pesticides.
I merely suggest that when we start to say that heterogeneous conditions like autism are 'related' to this month or that month of birth or conception, we start to enter some pretty shaky territory.

The paper which attracted my interest was this one published in Epidemiology from the MIND Institute. It has been followed by quite a few headlines like this one. Their findings: based on an analysis of quite a large participant base of people with autism (n=19,238), there was an increased 'risk' of autism where conception occurred in December through to March.

I say an increased risk but the odds ratios (OR) ranged between 1.08 - 1.16). Assuming that 1 implies the likelihood of something occurring in equal measure across 2 groups, you can see that the stats are not exactly jumping out at us (even with the confidence interval data presented). Consider as a comparator the OR of developing autism and living close to a motorway (freeway) by some of the same author group (OR = 1.86, CI: 1.04-3.45). See what I mean?

The MIND paper is not the first (and no doubt won't be the last) to suggest such a correlation to onset of autism. Last year we had this paper which suggested that in a UK cohort, conception during the summer months was associated with an increased risk of autism. Yes, I know that the participant group was far, far less (n=86) but importantly this was a prospective study based on the ALSPAC initiative.

I could go on with other studies, all seemingly suggesting different things. There are of course various factors to consider: where the study was conducted, how many people were included, etc. All of which can (and do) impact on the results obtained. There may very well be some sub-groups of autism with a seasonality factor attached, but at the moment we can't say for sure.

Importantly also some studies have reported no seasonality effect. This paper from Fred Volkmar and colleagues says it all in their title - even showing no association when sub-categorising participants based on language ability.

I will close with our contribution to this issue published in 2009. We did a slightly different kind of analysis, looking at the various sub-diagnoses of pervasive developmental disorder according to season of birth compared with population data (from the ONS) and found... absolutely nothing - no differences across the different cohorts and no differences overall. I would not presume to say that this issue is closed but I am inclined to say whilst an interesting area of research there is probably no overall relationship with autism.